Abstract
In periodic sequences of pseudo-random numbers generated by multiplicative congruential schemes, terms at certain critical distances are strongly correlated. For powers of two moduli a fast arithmetic method for computing these distances is given and applied to several generators. For all of them the length of the sequence that can be safety used turns out to be much shorter than the period. These correlations should be taken into account in parallel computations (e.g., Monte Carlo computations) when a single pseudo-random sequence is partitioned among concurrent processors.
Citation Formats
De Matteis, A, and Pagnutti, S.
Critical distances in pseudo-random sequences generated with composite moduli.
Italy: N. p.,
1992.
Web.
De Matteis, A, & Pagnutti, S.
Critical distances in pseudo-random sequences generated with composite moduli.
Italy.
De Matteis, A, and Pagnutti, S.
1992.
"Critical distances in pseudo-random sequences generated with composite moduli."
Italy.
@misc{etde_10139014,
title = {Critical distances in pseudo-random sequences generated with composite moduli}
author = {De Matteis, A, and Pagnutti, S}
abstractNote = {In periodic sequences of pseudo-random numbers generated by multiplicative congruential schemes, terms at certain critical distances are strongly correlated. For powers of two moduli a fast arithmetic method for computing these distances is given and applied to several generators. For all of them the length of the sequence that can be safety used turns out to be much shorter than the period. These correlations should be taken into account in parallel computations (e.g., Monte Carlo computations) when a single pseudo-random sequence is partitioned among concurrent processors.}
place = {Italy}
year = {1992}
month = {Dec}
}
title = {Critical distances in pseudo-random sequences generated with composite moduli}
author = {De Matteis, A, and Pagnutti, S}
abstractNote = {In periodic sequences of pseudo-random numbers generated by multiplicative congruential schemes, terms at certain critical distances are strongly correlated. For powers of two moduli a fast arithmetic method for computing these distances is given and applied to several generators. For all of them the length of the sequence that can be safety used turns out to be much shorter than the period. These correlations should be taken into account in parallel computations (e.g., Monte Carlo computations) when a single pseudo-random sequence is partitioned among concurrent processors.}
place = {Italy}
year = {1992}
month = {Dec}
}